In
d
o
n
e
sian
Jou
r
n
al of
Ele
c
tr
i
c
a
l
En
g
in
e
erin
g
a
n
d
C
om
pu
ter S
c
ien
ce
Vol.
14, No.
1, April 2019,
pp.
210~218
ISSN: 2502-
4752,
DOI
:
10.115
91/ijeecs.
v
14.
i
1
.
pp210-218
210
Jou
rn
a
l
h
o
me
pa
ge
:
ht
tp:
//i
a
e
score
.
com
/
j
o
u
r
na
l
s
/
i
n
d
e
x
.
p
hp/
i
j
eec
s
Cervical can
cer detection method
usi
n
g
an impr
ove
d cellular neura
l netw
o
rk (C
N
N) algorithm
Azi
a
n Aza
m
imi
Abdu
l
l
a
h
1
,
A
a
fion
F
on
e
t
ta D
i
c
k
s
on
Gion
g
2
,
Nik
Adi
l
a
h
H
a
n
i
n
Z
a
h
r
i
3
1,
2
S
c
ho
ol
o
f M
echat
ron
i
c E
n
g
i
neeri
n
g
,
Univ
e
rsiti
M
a
lays
ia Perlis
,
Ma
la
ysi
a
3
S
c
ho
ol
o
f
Co
mpu
t
er
a
nd
C
ommu
ni
ca
t
i
o
n
E
ngin
e
ering
,
U
ni
versi
t
i M
a
lay
s
i
a
P
erli
s, M
al
ays
i
a
Art
i
cl
e In
fo
ABSTRACT
A
r
tic
le hist
o
r
y
:
R
e
c
e
i
v
e
d
Sep
2
5
,
2
018
Re
vise
d N
ov
26,
201
8
A
c
c
e
pte
d
D
ec 8,
201
8
Cervi
cal
can
c
e
r
is
t
he
s
econ
d
m
o
s
t
comm
o
n
i
n
M
a
lays
ia
a
nd
th
e
f
ou
rth
f
r
equ
e
nt cancer am
o
ng wo
men
in
w
orl
d
w
i
de.
P
ap sm
ear
t
est
is
o
f
ten ig
nored
alt
h
o
ugh
it
is
act
uall
y
us
eful,
b
e
n
e
f
i
ci
a
l
a
nd
ess
e
nt
ial
as
s
cr
e
e
n
in
g
to
ol
f
or
cervi
cal
canc
e
r.
H
o
w
ev
er,
P
a
p
sm
ear
i
m
a
g
e
s
ha
v
e
l
ow
s
en
sitiv
ity
a
s
w
e
ll
a
s
sp
ecifi
c
ity
.
Th
eref
ore,
i
t
is
d
iffi
cul
t
t
o
de
t
e
rm
i
n
e
wh
eth
e
r
t
h
e
ab
norm
a
l
cells
are
can
cero
u
s
or
n
ot.
Recent
l
y,
c
om
p
u
t
e
r-based
a
lg
orith
m
s
a
re
w
idel
y
used
in
cervical
cancer
s
cre
e
ni
ng
.
In
t
his
st
udy,
a
n
im
prov
ed
cel
lul
ar
n
eural
net
w
o
r
k
(CN
N
)
alg
o
rit
h
m
is
p
ro
po
sed
as
t
he
s
olut
io
n
to
d
etect
t
h
e
can
cerous
cell
s
i
n
real-time
by
und
ergo
in
g
th
e
im
age
p
r
oc
ess
i
n
g
o
f
P
a
p
s
m
ear
i
m
a
ges
.
A
f
e
w
t
e
m
p
l
a
t
e
s
are
com
b
in
ed
a
n
d
m
o
d
ified
to
f
o
r
m
an
i
deal
C
N
N
al
gorith
m
to
d
etect
t
he
can
c
e
ro
us
cel
ls
i
n
tot
a
l
o
f
1
1
5
P
ap
s
mear
i
mag
e
s.
A
M
A
T
L
A
B
bas
e
d CN
N
i
s
d
evel
op
ed f
o
r
an
au
to
m
a
ted
det
ectio
n of cerv
i
x can
cerou
s cel
ls
wh
ere
t
h
e
tem
p
lates
s
e
gm
ent
e
d
th
e
nucl
e
us
o
f
t
h
e
cell
s
.
F
r
o
m
t
h
e
si
mula
ti
on
resu
lts,
ou
r
p
r
o
p
os
ed
C
NN
a
l
g
o
ri
thm
can
d
ete
c
t
t
h
e
cervi
x
c
a
nce
r
cells
aut
o
m
a
ti
call
y
wit
h m
o
re th
a
n 8
8
%
accuracy.
K
eyw
ord
s
:
C
e
llu
lar
neur
al
netw
o
r
k
C
e
rvi
c
al
c
an
cer
Im
age
proce
ssing
MATLAB
Pap
sm
ea
r
Co
pyri
gh
t © 2
019 In
stit
u
t
e
of Advanced
En
gi
neeri
n
g
an
d
Scien
ce.
All
rights
res
e
rv
ed.
Corres
pon
d
i
n
g
Au
th
or:
A
z
ian
A
z
a
m
im
i A
b
d
u
l
l
a
h
,
Sch
o
o
l
of
M
ech
a
t
ro
ni
c
Engi
nee
r
i
n
g
,
Uni
v
ersi
ti
M
al
ays
i
a P
e
rlis,
P
a
uh P
u
tra
Cam
pus,
0260
0
A
r
au,
P
e
rlis,
M
a
lay
s
ia.
Em
ail:
aza
mim
i
@
u
n
i
m
a
p.
ed
u
.
m
y
1.
I
N
TR
OD
U
C
TI
O
N
Cer
v
i
c
al
c
a
n
ce
r
i
s
a
c
a
n
cer
o
f
t
h
e
c
e
rv
ix,
whic
h
i
s
c
omm
onl
y
ca
u
s
ed
by
a
v
i
rus
na
m
e
d
H
u
m
a
n
Papi
llomavirus
(HPV)
[1].
T
h
e
v
ir
us
c
a
n
d
am
age
ce
lls
i
n
t
h
e
c
e
r
vix
,
n
ame
l
y
,
s
qu
a
m
ou
s
c
e
ll
s
a
n
d
gl
an
dul
ar
ce
l
l
s
tha
t
m
ay
d
eve
l
o
p
i
n
t
o
squ
a
m
ous
c
e
l
l
car
cinoma
(
canc
e
r
of
the
s
quam
o
us
c
e
l
l
s
)
a
nd
ade
noc
a
r
cin
o
m
a
(ca
n
ce
r
of
t
h
e
g
l
a
n
d
u
l
a
r
c
e
lls
),
r
espec
t
i
v
e
l
y.
B
ase
d
o
n
Wor
l
d
H
eal
t
h
O
rg
an
i
z
ati
o
n
(W
HO),
c
erv
i
c
a
l
c
a
n
c
e
r
i
s
the
fo
urt
h
m
ost
freq
u
e
n
t
ca
nce
r
i
n
w
o
m
e
n
afte
r
b
r
east,
c
olor
ect
al
,
an
d
lu
ng
c
a
n
ce
rs
w
i
t
h
a
n
est
i
ma
t
e
d
53
0,0
00
new
ca
ses
i
n
2
0
1
2
r
epre
sen
tin
g
7.
9%
o
f
a
l
l
fe
ma
le
c
a
n
ce
r
s
[
2]
.
A
pprox
ima
t
e
l
y
90
%
t
h
e
2
7
0
,00
0
dea
t
hs
f
r
o
m
c
e
r
vica
l
c
a
n
ce
r
in
2
0
1
5
occ
u
rred
i
n
l
ow
a
n
d
m
id
d
l
e-
in
co
me
c
o
unt
ri
es
[
3
]
.
M
e
anwh
il
e,
in
M
a
l
a
y
s
i
a
,
t
h
e
c
ervi
x
c
a
n
c
e
r
w
as
t
he
s
e
c
o
nd
c
o
m
m
on
ca
nc
er
a
mo
n
g
w
om
en,
be
hi
n
d
t
he
b
rea
s
t
c
a
n
c
e
r
[
4]
.
R
a
te
o
f
inc
i
de
nce
of
c
ervica
l
ca
ncer
i
ncre
as
ed
a
fter
r
ea
ch
i
n
g
3
0
y
e
a
r
s
o
f
a
g
e
a
n
d
p
e
a
k
s
f
r
o
m
6
5
-
6
9
y
e
a
r
s
ol
d.
I
n
Ma
la
y
s
ia
n
s
t
at
ist
i
c,
o
ut
o
f
a
l
l
t
h
e
ra
c
e
s,
I
ndia
n
w
o
m
e
n
p
o
ssess
t
he
h
i
g
hes
t
o
cc
urre
nce
su
bse
que
n
tly
beh
i
nd i
s
Ch
i
n
e
se foll
o
w
e
d b
y
Ma
l
a
y
. 10.
3
per
10
0,00
0
p
opu
la
ti
o
ns ar
e
t
he
A
S
R
f
or
I
ndian fem
ale
s
[5]
.
P
a
p
smear
screening
test
i
s
the
m
e
th
od use
d
to
l
o
o
k
for
pre
-ca
n
c
e
r
s
,
w
h
i
c
h
a
re
t
h
e
cell
s
t
ha
t
ch
ang
e
d
on
t
h
e
c
e
rv
ix
t
ha
t
m
i
gh
t
be
c
o
me
c
e
r
vica
l
c
a
nc
er
i
f
the
y
a
re
n
o
t
t
reat
e
d
a
ppropr
i
a
tely.
The
screening
f
o
r
ce
rvica
l
c
a
n
cer
w
i
l
l
de
tect
p
re
cursor
l
e
s
i
o
ns
t
her
e
fore
a
l
l
o
w
e
a
r
l
y
a
nd
pot
en
ti
a
l
ly
l
e
s
s
i
nvas
i
v
e
t
reat
men
t
t
h
a
t
w
h
a
t
i
s
r
e
q
u
i
r
e
d
f
o
r
d
i
s
e
a
s
e
s
t
h
a
t
c
a
u
s
e
t
h
e
s
y
m
p
t
o
m
s
.
P
a
p
s
m
e
a
r
t
e
st
i
s
a
s
i
mpl
e
,
ef
f
ect
i
v
e
a
n
d
ch
ea
p
,
b
ut
i
t
is
n
o
t
a
d
ise
a
s
e
spec
ific t
est,
w
hi
c
h
ha
s
l
o
w
sens
i
t
i
vi
t
y
an
d
s
pec
i
fici
t
y
[
6]. A
no
ther
d
i
s
a
d
v
a
nt
a
g
e
o
f
P
ap sm
e
a
r
is
t
ha
t
a
n
a
bn
o
r
m
a
l
P
a
p
sme
a
r
result
doe
s
n
o
t
a
l
w
a
ys
i
nd
i
c
a
t
e
c
a
n
c
e
r
.
C
e
l
l
s
so
met
i
m
e
s
a
p
p
ea
r
abn
o
r
m
al
b
u
t
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
ones
i
a
n
J
E
lec
En
g & Co
mp
S
c
i
IS
S
N
: 2502-
47
52
Ce
rv
ica
l
c
a
nce
r
d
e
tec
t
i
o
n
m
e
t
h
o
d
usi
ng an i
m
prov
e
d
cel
lu
lar ne
ura
l
ne
t
w
ork
…
(Azia
n
Azam
im
i
A
b
d
u
llah)
21
1
the
y
a
re no
t
c
ance
ro
us. There
f
or
e
,
it
is
d
i
f
f
i
c
u
l
t
t
o
d
e
te
r
m
in
e
w
he
ther the
a
bn
orma
l
c
e
l
l
s are
c
a
nce
r
ous
o
r
no
t.
Wi
t
h
t
hese
l
i
m
it
a
t
i
o
ns,
t
h
e
ca
nce
r
o
u
s
ce
ll
s
w
ill
be
h
ard
to
d
e
t
ec
t
an
d
it
w
i
ll
take
a
l
o
nge
r
t
i
me
f
or
t
he
radi
ol
o
g
ist
or
doc
tor
a
nd
ot
her
me
dica
l
p
e
rso
nne
l
to
d
eterm
i
ne
t
h
e
ex
i
s
t
e
n
c
e
o
f
c
a
n
c
e
rou
s
cel
ls
a
nd
con
s
eq
ue
n
t
l
y
w
il
l
l
e
a
d
t
o
th
e
sprea
d
in
g
of
i
t
t
o
t
he
s
urrou
n
d
i
n
g
ce
lls.
Rec
e
n
t
l
y,
t
h
e
re
a
re
m
a
ny
me
t
h
o
d
s
ut
iliz
i
n
g
com
p
ut
e
r
a
lgor
it
hm
s
for
dete
c
t
i
n
g
the
c
a
nc
ero
u
s
ce
l
l
s
i
n
ce
rvi
x
[
7]–[1
2
]
.
T
hese
m
e
t
hods
e
mpl
o
ye
d
sever
a
l
ima
g
e
proce
ssi
ng
tec
h
n
i
que
s
w
i
t
h
c
lass
ifica
t
i
o
n
alg
o
ri
t
h
m
s
to
d
i
a
g
n
o
se
cer
vi
c
a
l
ca
ncer
.
Basical
l
y
,
there
w
e
re
f
o
u
r
sta
g
es
o
f
the
com
p
u
t
er
s
ys
tem
inc
l
u
d
i
n
g
im
age
e
nha
nce
m
e
n
t,
f
eature
s
e
x
t
r
acti
o
n,
fe
at
ure
se
lec
t
i
on,
a
nd
c
l
a
s
s
i
f
i
c
a
t
ion.
H
ow
ever
,
there
ar
e
st
i
l
l
l
i
m
i
t
a
t
ions
i
n
a
u
t
o
m
a
ted
de
te
c
tio
n
w
i
t
hou
t
un
derg
o
i
n
g
t
he
se
f
o
u
r
sta
g
es.
O
n
t
he
o
t
h
er
h
a
nd,
c
el
l
u
l
a
r
n
e
ura
l
ne
tw
ork
s
(
C
N
N
)
a
re
a
p
ara
l
l
e
l
c
o
m
p
ut
in
g
para
d
i
g
m
s
i
m
ilar
t
o
n
e
u
ra
l
ne
tw
orks,
w
i
t
h
t
he
d
i
f
fer
e
nce
tha
t
c
o
m
munica
tio
n
is
a
llow
e
d
b
e
t
w
e
e
n
nei
g
hb
o
u
ri
ng
un
its
o
n
l
y.
C
ell
u
lar
ne
ura
l
n
e
t
w
o
r
k
(
CN
N
)
i
s
a
ne
two
r
k,
w
hich
i
s
l
o
cal
l
y
c
o
n
n
e
ct
ed
[
13]
.
The
ou
t
p
u
t
o
f
t
h
e
ne
uro
n
i
s
c
o
n
n
ec
te
d
w
i
t
h
t
he
i
n
p
u
t
of
e
ve
ry
n
e
u
ron
in
3
x3
n
ei
gh
bo
rhoo
d
s
.
Li
k
e
wi
se,
the
i
n
p
u
t
s
of
n
eur
ons
a
re
c
on
ne
c
t
ed
o
nl
y
to
o
utp
u
t
o
f
e
ve
ry
n
e
u
ro
n
in
i
ts
3
x
3
n
ei
gh
bor
h
ood.
I
n
ima
g
e
proce
ssi
ng,
C
N
N
ba
sic
a
lly
w
il
l
d
o
t
he
m
app
i
ng
or
t
ran
s
forma
t
i
o
n
o
f
an
i
np
ut
i
mag
e
t
o
it
s
co
rre
s
p
ond
i
n
g
ou
tpu
t
i
ma
ge.
F
o
r
insta
n
ce,
a
ny
i
n
p
u
t
ima
g
es
i
n
a
n
a
l
o
g
u
e
f
orm
c
an
b
e
tra
n
sform
e
d
i
n
to
a
s
pe
c
i
fic
ou
t
put
ima
g
e,
w
h
i
ch
i
ts
v
a
l
ue
i
s
i
n
b
inar
y.
In
C
NN,
f
eed
ba
ck
c
o
n
n
ec
ti
o
n
s
are
p
r
ese
n
t
an
d
ea
ch
o
f
t
h
e
n
e
u
r
on
o
f
the
netw
ork
ap
p
l
i
e
d
t
h
e
sim
ila
r
p
r
oc
essi
ng
fu
nc
ti
o
n
.
Cel
l
ular
N
eura
l
N
e
tw
ork
uses
t
hese
t
h
r
ee
m
ain
o
p
er
ati
o
n
s
nam
e
ly,
fi
l
t
e
r
i
ng, se
g
me
n
t
a
t
i
on
an
d e
d
ge de
t
e
c
t
i
on.
CANDY
So
ft
ware
i
s
an
a
p
p
li
cat
i
on
d
evel
op
ment
a
nd
en
vi
ro
n
m
ent
t
o
olk
it
ba
se
d
CN
N
for
w
i
n
dow
s
[1
4].
It
i
s
als
o
k
no
w
n
a
s
V
i
s
u
a
l
M
o
u
se
S
o
f
tw
are
P
l
a
t
f
o
r
m
or
V
isM
ouse
[1
5].
Few
tem
p
la
te
s
w
e
r
e
exp
l
ored
a
n
d
cons
i
d
ere
d
s
uch
as
b
l
u
e
cha
n
ne
l
e
x
t
r
acti
o
n
te
mp
la
t
e
,
c
ont
r
a
st
e
nha
nce
m
e
n
t
te
mpla
t
e
,
me
dian
f
i
lter
t
e
m
p
lat
e
,
b
i
na
r
y
e
dge
d
e
t
ec
ti
on
t
em
pla
t
e
a
n
d
ho
l
l
ow
-co
n
c
a
v
e
tem
p
la
te.
C
N
N
is
c
h
o
se
n
for
the
me
tho
d
o
f
de
te
cti
o
n
i
n
m
ed
i
cal
i
m
a
ge
p
ro
ce
ssi
n
g
d
u
e
t
o
i
ts
e
ff
i
c
i
e
n
cy
i
n
pa
tt
e
r
n
re
c
o
g
n
iti
on
a
nd
i
ma
ge
proce
ssi
ng
[1
6
]
–[2
0
].
C
N
N
alg
o
ri
t
h
m
is
a
ls
o
a
p
p
l
ie
d
in
m
emrist
o
r
[2
1]–[2
3
],
s
r
obotic
s
[24]
,
d
i
t
r
ibu
t
e
d
netw
ork
[2
5]
a
nd
r
a
i
n
d
ro
p
detec
t
i
on
[26]
.
H
e
nce
,
i
n
th
i
s
s
t
u
d
y
,
a
n
e
w
d
etec
tio
n
m
e
th
o
d
by
us
i
n
g
CNN
a
l
go
rith
m wi
l
l
b
e
u
sed
t
o
d
et
ect
t
h
e
can
ce
ro
us ce
r
v
i
c
a
l
c
e
ll
s i
n
a
sh
or
t
e
r tim
e.
2.
RESEARCH
M
ETH
O
D
The
most
i
mp
orta
nt
p
a
r
t
in
t
hi
s
pro
j
ect
i
s
d
e
si
gni
ng
the
te
m
p
la
te
a
nd
d
e
v
el
o
p
me
nt
o
f
the
al
gor
ithm
to
d
e
t
ect
t
he
c
e
r
vi
x
canc
e
r
b
y
u
s
i
ng
P
a
p
s
m
ear
i
m
a
g
e
s.
P
rior
t
o
t
he
d
e
v
el
o
p
me
nt
o
f
the
i
m
pro
v
e
d
a
lgor
i
t
hm
,
t
h
e
c
h
a
r
ac
t
e
ri
st
i
c
s
o
f
e
ac
h
i
m
a
g
es
n
e
e
d
to
b
e
an
aly
z
e
d
a
n
d
C
NN
nee
d
t
o
be
u
n
d
er
st
o
o
d
f
u
l
l
y.
T
o
de
sign
a
t
e
mp
l
a
t
e
, t
h
e in
it
i
a
l
st
at
e, b
oun
d
a
ry
co
ndi
tion
, f
e
ed
b
a
ck
a
nd
t
hre
s
h
o
l
d
v
a
l
ue ha
s
t
o be
s
tud
i
e
d
b
efor
eha
nd.
A
to
ta
l
o
f
1
3
i
m
a
g
es
w
e
r
e
c
o
l
l
ec
te
d
from
H
U
S
M
K
uba
ng
K
e
r
i
a
n
a
n
d
a
no
th
e
r
1
02
i
ma
g
e
s
re
t
r
i
e
ved
f
r
o
m
P
a
p
s
mea
r
i
mag
e
d
a
t
ab
ase.
A
t
o
t
al
o
f
11
5
i
m
a
g
es
w
ere
u
s
e
d
t
o
s
i
m
ul
ate o
u
r
de
ve
l
o
ped MA
TLA
B based
C
N
N
s
i
m
u
l
a
t
o
r
.
T
h
i
s
M
A
T
L
A
B
b
a
s
e
d
C
N
N
s
i
m
u
l
a
t
o
r
i
s
a
u
s
e
r
-
f
r
i
e
n
d
l
y
s
o
f
t
w
a
r
e
.
T
h
e
o
u
t
p
u
t
i
m
a
g
e
w
i
l
l
b
e
show
n
a
t
t
he
e
nd
o
f
t
he
t
ra
nsien
t
s.
T
hi
s
i
m
pro
v
ed
M
A
T
L
A
B
CN
N
s
i
m
u
lator
is
u
sed
for
P
a
p
sm
ear
im
ages
proce
ssi
ng
a
n
d
to
d
e
t
ec
t
the
ca
nce
r
o
u
s
ce
l
l
s.
F
low
c
h
art
of
t
he
a
lg
ori
t
h
m
f
or
cer
vix
c
a
nc
er
cel
l
d
e
te
ctio
n
i
s
show
n in F
igur
e 1.
Th
e
Graph
i
cal
U
se
r
In
t
e
rf
ac
e,
w
h
i
ch
i
s
in
M
ATLAB
,
i
s
ut
il
i
z
ed
f
or
t
he
p
urp
o
se
o
f
de
ve
lo
pm
en
t
of
the
C
N
N
S
i
mula
t
o
r
to
d
e
t
e
c
t
t
h
e
cer
v
i
c
a
l
canc
e
r
in
t
he
P
ap
s
me
a
r
i
m
age.
T
hese
t
em
plat
es,
whi
c
h
ar
e
des
i
g
n
e
d
f
or
t
he
d
e
t
e
c
t
i
o
n
o
f
c
e
r
vica
l
c
a
n
ce
r,
a
re
d
eve
l
ope
d
to
r
e
duc
e
the
c
o
ns
um
ing
tim
e
to
o
bt
ain
t
h
e
results.
T
he
p
r
o
cess
of
u
sing
th
e
MA
TLA
B
ba
se
d
CN
N
S
i
m
u
la
t
o
r
i
s
eas
y
by
s
i
m
p
l
y
lo
ad
t
he
i
n
put
i
m
a
ge
,
run
the
im
age
a
nd
t
h
e
fina
l
s
i
m
u
late
d
ou
tpu
t
i
ma
ge
w
i
l
l
be
d
isp
la
ye
d
a
t
t
h
e
e
nd
of
t
he
t
e
m
plate
s
.
The
G
U
I
o
f
the im
pr
ove
d
M
a
tl
a
b
CN
N
si
mula
tor
is sh
o
w
n in F
ig
ure
2.
2.1.
Te
mp
late
D
esign
Te
mpla
te
1
i
s
a
B
l
ue
C
ha
nne
l
E
x
tra
c
t
i
o
n
tem
p
la
te
t
h
a
t
ca
n
b
e
f
o
u
nd
i
n
C
N
N
l
i
brar
y
[2
7].
The
t
e
mp
la
t
e
s
et c
on
ta
i
n
ed o
f
A (
f
ee
dbac
k
)
,
B
(
con
t
ro
l)
a
nd
Z (
bias)
t
e
mpl
a
t
e
a
n
d
t
he
y a
r
e
show
n as
b
el
ow
:
T
e
m
p
lat
e
A
=
T
em
plat
e
B
=
Bia
s
,
Z
=
-
0.7
The
tem
p
lat
e
i
s
app
l
i
e
d i
n
the
M
atl
a
b CN
N
sim
u
l
a
t
o
r
a
n
d
t
h
e r
e
s
ult
is
s
h
o
w
n as F
i
g
ure
3.
0
0
0
0
1
0
0
0
0
2
-
2
2
0
0
0
2
-
2
2
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2502-
4752
I
n
do
n
e
si
an
J
E
l
e
c
E
n
g
&
C
o
m
p
S
ci
, V
o
l
.
1
4
,
No. 1, April 2019 :
210 –
2
0
1
8
21
2
Figure
1.
Flow
char
t
of
C
NN
alg
o
ri
t
h
m
for
c
e
r
v
ix
canc
e
r
detec
t
i
on
F
i
gur
e 2.
G
U
I
of Im
prove
d
M
a
tl
a
b
CN
N
S
i
mulat
o
r.
(a
) O
r
i
g
ina
l
I
mage
(
b) S
imulat
e
d
Im
a
ge
F
i
gure
3.
O
rigi
nal an
d
si
m
u
la
t
e
d
i
m
age
us
in
g
Blue
Cha
n
n
e
l
Ex
t
ra
ct
ion
tem
p
l
a
te.
F
r
om
F
igure
3,
t
he
s
imu
l
ate
d
resul
t
c
a
n
n
o
t
s
how
th
e
a
c
t
ua
l
nuc
l
e
u
s
o
f
t
he cell.
T
o o
b
t
a
in the des
i
r
ed i
m
a
ge
,
a
modi
fic
a
t
i
on
o
n
t
he
t
e
m
p
l
a
t
e
is
m
a
d
e
an
d
is
s
how
n
be
l
o
w
.
T
he
m
o
d
i
fi
e
d
t
e
m
pla
t
e
p
r
o
duc
ed
t
he
r
esu
lt
show
n
in Fig
ure
4
.
T
e
m
p
late
A
=
T
e
m
plate
B
=
1
1
1
1
-
7
.
5
1
1
1
1
2
-
2
2
0
-
2
0
2
-
2
2
Sta
r
t
T
e
mp
lat
e
1: Modi
f
i
ed B
l
u
e
Ch
ann
e
l E
x
t
r
ac
t
i
o
n
T
em
p
l
ate
Tem
p
l
a
te
2
:
Modif
i
ed
C
ontrast
En
hancem
en
t
T
e
m
pla
t
e
Temp
l
a
te 3: I
m
prov
ed
H
ollow
Co
ncav
e Tem
p
la
t
e
End
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
ones
i
a
n
J
E
lec
En
g & Co
mp
S
c
i
IS
S
N
: 2502-
47
52
Ce
rv
ica
l
c
a
nce
r
d
e
tec
t
i
o
n
m
e
t
h
o
d
usi
ng an i
m
prov
e
d
cel
lu
lar ne
ura
l
ne
t
w
ork
…
(Azia
n
Azam
im
i
A
b
d
u
llah)
21
3
Bia
s
,
Z
=
1.5
(a) O
r
igina
l
im
a
ge
(
b) S
imula
t
e
d
i
ma
ge
F
i
gure
4.
O
rigina
l an
d
si
m
u
la
t
e
d im
age
us
in
g
modifie
d
B
l
u
e
Cha
n
ne
l Ex
tra
c
ti
o
n
tem
p
l
a
t
e.
F
i
gure
4
s
h
ow
s
th
a
t
t
he
m
o
d
i
f
ie
d
tem
p
la
t
e
i
s
able
t
o
fi
l
t
e
r
o
u
t
t
he
nuc
le
us
o
f
t
h
e
c
e
l
l
.
T
h
e
unw
an
t
e
d
bac
kgr
o
u
n
d
is
fi
l
t
er
e
d
ou
t
a
n
d
he
n
c
e
the
o
ut
p
u
t im
a
g
e is
m
o
r
e
p
r
e
senta
b
le for the de
t
ect
i
o
n
o
f
a
bn
orm
a
l cell
s
.
N
e
xt,
Te
mp
la
t
e
2
i
s
use
d
,
w
h
i
c
h
is
t
he C
ont
ra
st
E
nha
nc
em
ent
tem
pla
t
e
i
n
o
rde
r
t
o proce
e
d
w
i
t
h
the
n
e
x
t
s
t
e
p
.
The
t
e
mp
la
t
e
i
s show
n be
low
and
t
h
e
s
i
mula
te
d re
su
lt i
s
sh
o
w
n
i
n
F
i
gure
5.
Te
m
p
l
a
t
e
A
=
Te
m
p
l
a
t
e
B
=
Bia
s
,
Z
=
-
0.7
(a)
O
r
igi
n
a
l
i
mage
(b)
S
i
m
u
lat
e
d
im
age
F
i
gure
5.
O
rigina
l an
d
si
m
u
la
t
e
d im
age
us
in
g
Co
ntra
s
t
E
nha
nce
m
e
nt
t
em
pla
t
e.
F
r
om
F
igure
5,
t
he
s
imul
at
ed
r
e
s
ult
s
h
o
w
s
a
n
e
nha
nc
em
ent
of
t
he
d
a
r
k
e
s
t
a
r
e
a
o
f
t
h
e
o
r
i
g
i
n
a
l
i
m
a
g
e
.
H
o
w
e
ve
r,
t
he
u
n
n
ec
essar
y
b
a
c
kgr
ou
n
d
r
e
a
ppe
ars
a
nd
t
h
e
r
efore
the
t
em
pl
a
t
e
nee
d
s
to
b
e
m
o
dif
i
e
d
a
s
sh
ow
n
bel
o
w
:
Te
m
p
l
a
t
e
A
=
Te
m
p
l
a
t
e
B
=
B
i
as
,
Z=
0
.2
Af
t
e
r
appl
yin
g
t
h
e
m
od
ifi
e
d
Co
nt
rast
E
nh
an
ce
me
n
t
t
emp
l
a
t
e
,
i
t
i
s
o
b
ser
v
ed
t
ha
t
on
l
y
t
he
nuc
l
e
us
a
pp
e
a
re
d
in
the
si
m
u
la
te
d
ima
g
e
as
s
h
o
w
n
i
n
F
i
gure
6
.
B
y
a
d
j
u
s
tin
g
the
simu
la
ti
on
tim
e
an
d
t
h
e
tem
p
l
a
t
e
,
t
h
e
ou
t
p
u
t
ima
g
e
sh
ow
s
o
n
l
y
t
he
d
arke
s
t
a
r
ea
of
t
he
p
r
e
vi
o
u
s
ima
g
e.
F
or
t
h
e
fina
l
s
t
e
p
,
Tem
p
la
te
3
,
w
h
ic
h
i
s
t
he
H
oll
o
w
Conc
a
v
e
tem
p
la
t
e
,
i
s
u
sed i
n
or
d
er
t
o
ge
t t
h
e
fina
l r
e
sul
t
.
The
tem
pla
t
e is
s
how
n
be
low
:
0
0.
25
0
0
0
0.
2
0
0.
25
0
0
-1
0
-1
4
-
1
0
-1
0
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2502-
4752
I
n
do
n
e
si
an
J
E
l
e
c
E
n
g
&
C
o
m
p
S
ci
, V
o
l
.
1
4
,
No. 1, April 2019 :
210 –
2
0
1
8
21
4
Te
m
p
l
a
t
e
A
=
Te
m
p
l
a
t
e
B
=
Bia
s
,
Z
=
3.2
5
(
a
)
O
r
i
ginal
im
age
(b)
S
i
m
u
lated
i
ma
g
e
F
i
gure
6.
O
rigi
nal a
n
d
simu
la
t
e
d
i
m
age
us
in
g
m
odi
f
i
e
d
Co
n
t
r
a
st
Enha
nc
em
ent tem
p
la
te
The
ori
g
ina
l
t
e
m
p
l
a
t
e
a
n
d
t
h
e sim
u
l
a
te
d re
sult
from
Te
m
plat
e
3
a
r
e
show
n i
n
F
i
gur
e 7.
(a)
O
r
igina
l
image
(b)
S
i
mulated
i
m
a
ge
F
i
gure
7. O
r
i
gi
na
l
an
d sim
u
l
a
te
d im
age
usin
g
H
o
l
l
ow
C
o
n
ca
ve
t
e
m
pl
ate
The
si
m
u
late
d
ima
g
e
i
n
F
ig
ure
7
afte
r
app
l
ying
the
ori
g
i
n
a
l
H
ol
lo
w
Co
n
cav
e
t
e
mp
l
a
t
e
d
o
e
s
n
o
t
sho
w
t
he
d
e
s
i
r
abl
e
out
put
.
It
i
s
s
u
pp
os
ed
t
o
di
sp
la
y
a
sm
a
l
l
nuc
le
us
t
hat
ind
i
ca
t
e
s
t
h
e
a
b
n
o
rm
al
c
e
ll.
H
ence
,
and
i
m
prove
d
H
o
l
l
ow
C
onc
a
v
e
tem
p
la
t
e
i
s
de
ve
lo
pe
d
for
t
h
e
de
tect
io
n
o
f
a
b
n
or
mal
cell
a
s
show
n be
low
:
Te
m
p
l
a
t
e
A
=
Te
m
p
l
a
t
e
B
=
Bia
s
,
Z
=
3.2
5
The
o
u
tp
ut
o
f
the
ima
g
e
afte
r
app
l
yin
g
t
he
n
ew
m
o
d
i
fied
H
ol
low
C
o
n
c
a
v
e
te
m
p
l
a
te
i
s
s
h
o
w
n
i
n
F
i
gure
8.
0
0
0
0
2
0
0
0
0
0.
5
0.
5
0.5
0.
5
2
0.5
0.
5
0.
5
0.5
0
0
0
0
5
0
0
0
0
0.
75
0.75
0
.75
0.
75
5
0.75
0.
75
0.75
0
.75
Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesia
n
J
Elec Eng
&
C
o
m
p
S
ci
ISSN:
2502-
4752
Cerv
ica
l
ca
nce
r
detec
t
i
o
n
m
e
tho
d
us
i
n
g
a
n
i
m
prove
d
ce
llu
l
a
r ne
ura
l
ne
tw
ork
…
(
A
zia
n
Azam
im
i
A
bdu
ll
ah)
21
5
(a) Orig
in
al imag
e
(
b
)
Simu
lat
e
d
imag
e
F
i
gur
e
8.
O
r
i
gi
nal
an
d
si
m
u
la
t
e
d
i
m
age
us
in
g
m
odi
f
i
e
d
H
ol
l
o
w
Co
n
c
av
e t
e
mp
l
a
t
e
F
i
g
u
r
e
8
sho
w
s
a
cl
e
a
n
i
m
a
g
e
wi
thou
t
unn
ece
ssa
ry
b
a
c
kg
r
o
und
a
n
d
n
oi
ses.
T
h
e
f
i
n
a
l
out
put
a
f
t
e
r
a
pp
ly
ing
the
mod
i
fie
d
H
ol
low
C
o
nc
a
v
e
t
e
m
p
la
t
e
s
how
s
the
sma
l
l
b
l
a
c
k
nuc
l
eu
s th
a
t
re
p
res
e
n
t
s
t
h
e
ab
no
rmal
c
el
l
.
3.
RESULTS
A
ND
A
NAL
YS
IS
From
this
s
tu
d
y
,
t
h
e
in
for
m
a
tio
n
o
b
ta
i
n
ed
f
r
o
m
t
h
e
s
i
m
u
la
tio
n
i
s
th
e
p
r
e
limi
n
a
r
y
resu
lt
,
wh
i
c
h
c
a
n
be
u
se
d
by
t
he
m
e
d
i
c
a
l
p
e
r
so
n
n
e
l
f
or
r
ec
omm
e
nda
ti
on
a
nd
f
i
na
l
d
ec
isi
o
n
r
e
gar
d
in
g
f
u
r
t
her
t
r
eatm
e
nt
o
f
the
pa
tie
nt
s,
i
f
a
n
a
bnor
ma
l
r
e
su
l
t
i
s
ob
ta
ine
d
.
Me
dica
l
per
s
o
nne
l
can
d
e
c
id
e
wh
et
h
e
r
th
e
ce
l
l
s
a
r
e
c
a
n
c
ero
u
s
o
r
no
n-
c
a
nce
r
o
u
s.
T
he
M
A
TLA
B
base
d
CN
N
ca
n
a
l
s
o
b
e
us
ed
b
y
t
h
e
me
d
ic
al
p
r
a
c
t
i
t
i
o
ner
s
o
r
s
t
ude
n
t
s
a
s
t
he
i
r
gu
i
d
el
i
n
es
i
n
d
e
te
r
m
i
n
in
g
c
a
n
c
e
r
ous
c
el
l
s
a
n
d
a
s
a
l
e
ar
ni
n
g
pur
p
o
se
.
I
n
t
e
rpre
tin
g
t
h
e
im
a
g
es
b
e
cam
e
easier
a
n
d
ca
n
r
e
d
u
ce
t
he
c
on
sum
e
d
t
i
m
e
.
Th
e
c
a
n
c
e
r
ou
s ce
l
l
s
a
r
e
in
vin
c
i
b
l
e
th
r
o
ugh
out
t
he
l
if
e
unt
il
w
o
m
e
n f
i
na
ll
y
d
ecide
o
n P
a
p S
m
ear
Test.
To
d
e
c
i
de
o
n
a
n
a
bn
or
m
a
l
or
nor
m
a
l
c
e
l
l
,
t
hi
s
au
t
o
ma
t
e
d
ce
r
v
i
x
ca
ncer
c
e
l
l
s
d
ete
c
t
i
o
n
whi
c
h
imp
l
e
m
enti
n
g
the
CN
N
alg
o
r
i
t
h
m
can
b
e
use
d
.
The
t
e
mpla
t
e
s
w
e
r
e
m
od
if
ied
a
nd
i
m
pr
o
v
e
d
i
n
or
de
r
to
b
e
in
s
y
n
c
w
i
t
h
t
he
P
a
p
sm
ea
r
im
a
g
es
g
i
v
e
n
.
The
ob
se
rva
t
i
o
n
i
s
b
ase
d
o
n
t
h
e
size
o
f
t
he
nuc
leu
s
a
nd
the
n
u
m
b
er
o
f
m
u
lt
ip
l
i
e
d
c
e
lls
i
n
a
n
i
m
a
ge
.
A
s
f
or
t
h
i
s
st
ud
y,
t
he
nuc
l
e
us
i
s
se
t
a
s
t
he
p
ara
m
e
t
ers
to
b
e
o
b
se
rv
ed
,
and
aft
e
r
t
h
e
simu
lat
i
on,
t
h
e
w
hite
a
r
e
a
r
e
pr
e
s
ents
t
he
nuc
leu
s
a
nd
b
l
ac
k
ar
e
a r
e
pr
e
s
ents
t
he
r
est
of
t
he
c
ell.
The
sim
u
la
t
i
o
n
result
in
d
e
t
e
c
ti
n
g
c
ance
ro
u
s
c
ells
i
n
P
a
p
sm
ea
r
i
m
a
ge
a
r
e
s
how
n
i
n
F
i
gur
e
9
a
n
d
F
i
gur
e 1
0
f
or
t
he
e
xam
p
l
e
of
abn
o
r
m
a
l
ce
lls,
r
e
spe
c
t
i
v
e
l
y.
F
igu
re
11
a
n
d
Fi
g
u
r
e
12
s
how
t
he
s
i
m
ul
ati
o
n
re
sult
o
f
P
a
p
s
m
e
a
r
i
m
a
g
e
f
o
r
t
h
e
e
x
a
m
p
l
e
o
f
n
o
r
m
a
l
c
e
l
l
s
.
I
t
i
s
o
b
s
e
r
ve
d
t
h
a
t
f
or
a
b
n
o
r
m
al
c
e
l
l
s
,
the
b
l
ac
k
d
o
t
in
d
i
cat
i
n
g
t
h
e
nuc
l
e
us
w
i
l
l
a
ppea
r
a
t
t
h
e
f
i
na
l
sta
g
e
.
U
n
l
i
k
e
f
o
r
n
o
r
m
a
l
c
e
l
l
s
,
t
h
e
b
l
a
c
k
d
o
t
(
n
u
c
l
e
u
s
)
w
i
l
l
di
sap
p
ea
r a
f
ter
Tem
p
late
3
.
F
i
gur
e
1
0
.
A
bnor
ma
l
ce
ll
s
o
f
P
ap
s
m
ear
i
m
a
ge
(
S
a
m
p
le
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2502-
4752
I
n
do
n
e
si
an
J
E
l
e
c
E
n
g
&
C
o
m
p
S
ci
, V
o
l
.
1
4
,
No. 1, April 2019 :
210 –
2
0
1
8
21
6
F
i
gur
e 1
1
.
A
bnorm
a
l
cells o
f
P
a
p sm
ea
r
image
(S
ample
2)
F
i
gur
e 1
2
.
N
o
rma
l
c
e
l
ls o
f P
a
p sm
ear im
a
ge
(
S
a
m
p
l
e
3)
Figure
1
3
.
Norm
al
cells o
f
Pap sm
ear ima
ge
(
Sample
4
).
The
pe
rce
n
t
a
g
e
f
or
e
a
c
h
ty
pe
o
f
ab
n
o
rm
al
c
ell
s
i
s
s
how
n
i
n
T
a
b
l
e
1
w
h
i
l
e
T
a
b
l
e
2
s
h
o
w
s
t
h
e
p
e
r
c
e
n
t
a
g
e
o
f
ac
cura
cy f
or e
ac
h
t
y
p
e
of t
h
e norm
a
l c
e
l
l
s.
F
rom
Table
1, i
t ca
n be co
n
c
l
u
d
e
d
t
h
a
t the
ov
era
l
l acc
urac
y for
t
h
e
65
of
a
b
norm
a
l
i
m
age
s
w
it
h
59
de
tecte
d
a
b
norm
a
l
i
t
i
e
s
i
s
90.7
7
%
.
T
he
a
ccur
acy
i
s
co
n
s
i
d
ere
d
h
i
gh
for
the
detec
t
i
o
n
o
f
a
b
norm
a
l
c
e
r
v
ica
l
c
e
l
l
s
.
From
T
a
b
l
e
2
,
it
is
o
b
s
er
ved
t
h
at
4
3
no
n-
canc
e
ro
us
P
ap
s
m
e
a
r
i
m
a
ges
are
cor
r
ec
t
l
y
d
e
t
e
ct
ed
o
ve
r
t
o
ta
l
o
f
5
0
im
age
s
a
n
d
t
he
a
cc
urac
y
i
s
85.
54
%.
T
he
limi
t
ed
num
ber
o
f
n
o
n
-
ca
ncer
ous
i
ma
g
e
s is t
he
m
a
i
n
fa
ctor why
t
h
e
a
ccur
acy
i
s
qu
i
t
e
l
ow
c
ompa
red
t
o
t
h
e
c
ance
rous im
a
ges.
Tab
l
e
1.
P
ercent
a
ge
o
f a
ccur
acy
f
or
eac
h c
e
l
l
(
Cance
r
ous)
Ty
p
e
o
f
cell
s
N
u
m
b
e
r
o
f
d
et
ect
ed
c
e
l
l
s
T
o
t
al
cell
s
s
i
m
u
l
a
t
ed
P
e
rc
e
n
ta
ge
o
f
a
c
c
u
r
a
c
y
(
%
)
L
i
ght
D
y
s
pla
s
tic
1
6
18
88.
89
M
ode
r
a
t
e
D
ys
pl
a
s
tic
1
5
16
93.
75
Se
v
e
r
e
D
y
s
pl
a
s
ti
c
12
13
92.
31
Ca
r
c
ino
m
a
In
S
it
u
1
6
18
88.
89
T
o
t
a
l
5
9
65
90.
77
Tab
l
e
2.
P
ercent
a
ge
o
f a
ccur
acy
f
or
eac
h c
e
l
l
(
N
on-c
a
nc
ero
u
s)
To
t
a
l
4
3
5
0
8
5
.
5
4
Ty
p
e
o
f
cell
s
Nu
m
b
er
o
f
un
d
e
t
e
ct
e
d
cel
ls
T
o
tal
ce
lls
si
m
u
lated
P
e
rc
ent
a
g
e
o
f
a
c
cu
r
a
cy
(
%
)
Norm
a
l
Supe
r
f
ic
ia
l
19
2
2
86.
36
Norm
a
l
In
t
e
r
m
e
d
ia
te
14
1
5
93.
33
Norm
a
l
Co
l
u
m
n
a
r
10
1
3
76.
92
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
ones
i
a
n
J
E
lec
En
g & Co
mp
S
c
i
IS
S
N
: 2502-
47
52
Ce
rv
ica
l
c
a
nce
r
d
e
tec
t
i
o
n
m
e
t
h
o
d
usi
ng an i
m
prov
e
d
cel
lu
lar ne
ura
l
ne
t
w
ork
…
(Azia
n
Azam
im
i
A
b
d
u
llah)
21
7
4.
CONCL
U
S
ION
CN
N
is
freq
u
e
n
tl
y
use
d
i
n
im
a
g
e
proc
ess
i
ng
,
hi
gh-s
p
ee
d
si
gna
l
p
roc
e
s
s
in
g,
v
ide
o
s
i
g
nal
p
r
oces
si
n
g
,
bi
olo
g
i
ca
l
vis
i
ons
a
nd
p
a
t
tern
r
e
c
o
g
n
it
io
n.
C
N
N
i
s
i
mp
le
m
e
nte
d
i
n
t
h
i
s
st
ud
y
part
icu
l
a
r
l
y
t
o
de
tec
t
t
h
e
cer
vix
ca
ncer
c
ell
s
a
u
t
oma
tica
lly
by
des
i
g
n
i
n
g
t
he
t
em
plat
es
a
n
d
f
ul
l
y
u
ti
l
i
z
i
ng
t
h
e
f
u
nc
tio
n
o
f
M
A
TLA
B
s
o
ft
w
a
re
.
Th
e
t
e
mpl
a
t
e
s
a
r
e
d
e
si
gn
e
d
a
ft
er
c
on
si
d
e
ring
e
ssenti
a
l
p
a
ra
me
t
e
r
s
s
uc
h
a
s
i
n
itia
l
s
t
a
t
e
,
b
o
u
n
da
r
y
c
on
d
i
t
i
on,
pi
xe
l
val
u
e,
c
on
t
r
o
l
t
em
pl
at
e,
f
ee
dbac
k
t
e
m
plate
an
d
thr
e
sh
ol
d
val
u
e.
B
a
s
e
d
o
n
the
sim
u
l
a
t
i
on
re
s
u
lts,
5
9
abn
o
rm
al
i
ma
ge
s
we
re
d
etec
te
d
corre
ct
ly,
o
u
t
of t
o
t
a
l
6
5
im
ages
.
Th
e
o
b
t
a
in
e
d
a
ccu
ra
cy
w
as
9
0
.
77
%
.
F
o
r
t
he
no
n-ca
ncer
o
u
s
im
ages,
43
i
m
ages
w
e
r
e
d
e
te
ct
ed
a
s
n
o
rm
al
cel
ls,
ou
t
of
t
o
t
al
5
0
im
a
g
es
w
i
t
h
ac
c
u
ra
cy
o
f
85.5
4
%.
F
or
f
ut
ure
improve
me
nt
,
m
o
re
i
m
a
ges
nee
d
t
o
be
s
imula
t
e
d
un
der
t
h
is
M
A
TLA
B
ba
se
d
CN
N
si
m
u
lator
for
m
o
re
a
ccu
r
ate
and
precise
res
u
lts.
Lac
k
o
f
i
m
age
s
h
i
n
d
e
r
s
t
he
p
re
cis
i
on
o
f
t
he
d
es
i
gne
d
tem
p
l
a
te.
E
v
e
r
y
ima
g
e
has
di
ffe
ren
t
c
om
p
o
sit
i
on,
b
ri
gh
tne
s
s
a
n
d
i
nt
en
si
ty
.
In
o
rd
er
t
o
ful
f
i
l
l
an
d
achi
e
ve
t
h
e
st
a
b
il
it
y
of
t
he
n
e
w
ly
d
e
s
ig
ne
d
tem
p
late
s,
m
ore
i
m
age
s
n
e
e
d
ed
f
or
t
he
s
i
m
ul
a
t
i
on
a
nd
a
n
a
l
ys
is.
In
t
he
f
u
t
ure
,
class
i
fica
t
i
o
n
c
an
b
e
m
a
de
w
her
e
t
he
c
a
l
c
u
l
a
t
i
o
n
o
f
th
e
si
z
e
o
f
t
he
n
uc
le
us
c
a
n
b
e
prec
ede
d
f
or
a
h
igh
ac
cura
cy a
n
d
p
re
cision.
ACKNOW
LEDG
E
MEN
T
S
Th
is w
or
k w
a
s fu
n
d
ed
b
y U
n
i
v
ersi
ti
Mala
ys
i
a
P
e
r
l
i
s
(U
niMA
P
)
res
e
a
rc
h gra
n
t
9
0
0
9
-0
00
53.
REFE
RENCES
[1]
Y.
J
u
s
man,
S
.
C.
N
g,
a
nd
N
.
A.
A
bu
O
s
m
an,
“Inte
lligent
sc
re
en
in
g
sys
t
e
m
s
f
o
r
cervi
ca
l
can
cer,”
S
cien
tif
ic
Wo
rld
Jo
urnal,
vol
.
20
14
.
2
0
1
4
.
[2]
J
.
F
er
l
a
y,
I
.
Soe
r
j
o
m
a
t
a
ram,
R
.
Dikshit,
S
.
Es
er
,
C
.
M
at
her
s,
M
.
Rebel
o
,
D
.
M
.
P
a
rki
n
,
D.
F
orm
a
n,
a
nd
F
.
Bray
,
“
C
ancer i
ncid
ence an
d m
o
rtality
w
o
rld
w
i
d
e,” IARC CancerBase
N
o.
1
1
G
L
O
BO
CAN 20
1
2
, 20
1
3
. .
[3]
W
HO
and
ICO,
“
Hu
ma
n
Papillomavi
rus
and
Rel
a
ted
Diseases
R
ep
ort
-
WORL
D,
”
HP
V
Inf
.
C
en
t.,
no.
Al
bani
a,
p
p.
1
–
1
38
,
2
01
4.
[4]
S
. Zarid
ah, “A rev
i
ew
o
f
cerv
i
cal
c
an
cer resear
ch
i
n M
a
lays
ia,
”
M
ed.
J
.
M
a
l
a
ysi
a
, v
ol
.
69
,
p
p
.
33
–4
1
,
2
01
4.
[5]
ICO H
P
V In
f
o
r
m
ation Centre,
“Human Pap
illomavi
rus
and Rela
t
e
d
Di
s
eas
es Report WORL
D,”
2
0
17.
[6
]
E
.
N
joro
ge,
S
.
R
.
Alty
,
M
.
R
.
Ga
ni,
an
d
M
.
A
lka
tib,
“Clas
s
i
fi
cati
o
n
o
f
cervi
cal
cancer
cel
ls
u
si
ng
F
T
I
R
dat
a
.,”
Ann
u
. In
t
.
Con
f
.
IEEE Eng
. Me
d
. B
i
o
l.
S
o
c
.
,
v
o
l
.
1
,
p
p.
5
33
8–
41
,
200
6.
[
7
]
G.
R
ig
on
,
C
.
V
a
l
lo
ne
,
A
.
S
ta
r
i
ta
,
M
.
F
l
a
vi
o
,
M
.
Vi
s
m
a
r
a
,
P
.
I
a
l
o
n
go
,
L
.
P
uti
g
n
a
ni
,
an
d
F
.
S
i
g
n
o
re
,
“
D
iag
n
o
s
tic
Accu
rac
y
o
f
M
R
I
in P
rim
a
ry
C
erv
i
cal
Cancer,” Open
J.
R
ad
io
l.
, v
o
l
.
2
,
pp
.
14
–2
1,
2
0
12.
[8]
S.
N
.
Su
l
a
i
m
an,
N.
A
.
Mat-Is
a,
N
.
H
.
O
thman,
a
nd
F
.
A
h
m
a
d,
“
I
m
p
ro
vem
e
nt
o
f
featu
r
es
e
x
t
ract
ion
p
r
oces
s
an
d
clas
sificati
on
o
f
C
ervical
cancer
f
o
r
t
h
e
N
e
u
r
a
l
P
ap
s
y
s
t
e
m
,
”
i
n
P
r
o
c
e
d
ia
C
o
m
pu
te
r
Sc
ie
n
c
e
,
2
01
5,
v
o
l
.
60
,
no
.
1
,
p
p.
7
5
0
–
75
9.
[
9
]
M
.
Z
h
a
o
,
A
.
W
u
,
J
.
S
o
n
g
,
X
.
S
u
n
,
a
n
d
N
.
D
o
n
g
,
“
A
u
t
o
m
a
t
i
c
s
c
reenin
g
of
c
erv
i
cal
cell
s
u
sing
b
lo
ck
i
m
a
ge
p
r
ocessi
n
g
,” B
i
o
med
. Eng
.
On
lin
e, v
ol
.
1
5
, no
. 1
,
20
16
.
[10
]
P
.
S
u
k
u
m
a
r
an
d
R.
K
.
Gnan
am
urth
y
,
“
Co
m
puter
a
id
e
d
s
cree
n
i
n
g
o
f
cervi
cal
c
an
cer
u
s
i
n
g
r
an
do
m
f
o
res
t
classifier,” Res.
J
.
Ph
arm.
B
i
o
l.
C
hem.
S
ci.
,
v
o
l
. 7
, no
. 1
,
2
0
16.
[
1
1
]
N
.
A
.
Ob
u
k
h
o
v
a,
A
.
A.
M
ot
yk
o
,
U.
K
ang
,
S.
-
J
.
Bae,
a
nd
D.
-
S
. Lee,
“
Aut
o
mated
i
m
age
an
al
ysis in
multispect
ral
sy
st
e
m
f
or
cerv
i
cal
c
a
n
cer
d
i
a
gn
ostic,
”
C
onf.
O
pen
In
no
v.
Ass
oc
.
F
r
u
c
t,
vol.
201
7
–
Ap
ril
,
2
0
1
7
.
[12]
L
.
W
e
i
,
Q
.
Gan,
a
nd
T
.
Ji
,
“Cerv
i
c
a
l
c
a
ncer
h
istology
imag
e
i
d
en
tif
i
cati
o
n
m
e
t
h
o
d
b
as
ed
on
t
e
xtu
r
e
an
d
les
i
o
n
area f
eature
s
,” Compu
t.
A
ssist
.
S
u
rg
.,
v
o
l
.
2
2
, p
p
.
18
6
–
1
9
9
,
2
0
17.
[13
]
L
.
O.
C
hu
a
and
L
.
Y
an
g,
“
Ce
l
l
ular
n
eural
n
e
tw
orks
:
t
h
eo
ry,
”
I
E
E
E
T
r
a
n
s
.
C
i
r
c
u
i
t
s
S
y
s
t
.
,
v
o
l
.
3
5
,
n
o
.
1
0
,
p
p
.
12
57
–12
72
,
1
988.
[14
]
L
.
O
.
C
hu
a
and
T.
R
os
ka,
Cellu
lar
n
e
ural
n
et
work
s
an
d
vis
u
al
c
om
p
u
ti
ng:
f
o
u
n
d
a
t
ion
an
d
ap
plicati
o
n
s
.
200
2
.
[1
5]
P
.
Szo
l
g
a
y,
A
.
ZarAnd
y
,
S
.
Zold
,
T
.
R
oska
,
P.
F
o
l
d
e
sy,
L.
Kek,
T
.
K
o
zek
,
K
.
L
asz
l
o,
I
.
P
e
t
r
as,
C.
R
ek
ec
zky,
I
.
S
zatm
a
ri,
an
d
D
.
B
alya,
“
T
he
c
o
m
p
u
t
a
ti
on
a
l
i
nf
rastru
ctu
r
e
for
c
ell
u
l
a
r
vis
u
a
l
m
i
c
roprocess
o
rs
,”
P
roc.
7
t
h
I
nt.
Con
f
.
M
i
croel
ectro
n.
N
eural
,
F
uzzy
B
io
-Ins
p
i
r
ed
S
yst.
M
icro
Neuro
19
99
,
p
p.
5
4–
60
,
1999
.
[16
]
A
.
A
.
A
bd
u
llah
and
H.
M
oh
am
a
d
d
i
ah,
“
D
evel
opm
ent
of
cell
u
l
a
r
neural
n
et
work
a
lg
orithm
f
o
r
det
ectin
g
l
u
n
g
canc
e
r
s
y
m
p
t
o
ms
,”
B
iom
e
di
cal
E
ngi
neeri
n
g
and
S
c
ien
ces
I
E
C
BE
S
20
10
I
EE
E
E
M
BS
C
onferen
ce
on
.
IE
EE,
pp.
1
38–
14
3,
201
0.
[1
7]
A
.
A.
A
b
d
u
l
l
a
h
,
B
.
S.
C
h
i
ze,
a
n
d
Y
.
Ni
sh
io
,
“I
mpl
e
men
t
ati
o
n
o
f
a
n
i
m
pro
v
ed
cell
u
lar
neu
r
a
l
n
etwo
rk
a
lg
orit
h
m
f
o
r b
r
a
i
n
tu
m
o
r d
e
tecti
o
n
,
” i
n
2
0
1
2 In
ternatio
nal
Con
f
eren
ce on
B
io
m
e
d
i
c
a
l
E
ngi
n
eeri
ng
(ICoBE),
20
12,
p
p
.
6
11–
6
1
5.
[1
8]
S
.
C.
L
in
g
,
A
.
A
.
A
bd
ul
la
h,
a
n
d
W.
K
.
W.
A
h
m
a
d
,
“De
s
ig
n
o
f
a
n
au
to
mat
e
d
breas
t
cancer
m
as
ses
d
e
tecti
o
n
i
n
mammog
r
am u
sin
g
Cell
u
lar
Neu
r
al
Network
(CNN) algo
r
i
t
h
m
,”
A
dv
.
Sci.
Let
t
.
,
vo
l
.
20
,
n
o
. 1
, p
p
. 2
54
–2
58
,
2
0
1
4
.
[19
]
R
.
Rou
h
i,
M
.
J
a
fari,
S.
K
asaei
,
an
d
P
.
K
esh
a
varzi
a
n,
“
Beni
g
n
a
n
d
m
al
ign
a
n
t
b
reast
t
u
m
o
rs
c
l
a
ssifi
catio
n
bas
e
d
on
r
egi
o
n
gro
w
ing
and
CNN
seg
m
en
tation
,
” E
xpert
S
y
s
t.
A
p
p
l.
,
vo
l
.
4
2,
n
o.
3
,
pp.
99
0
–
1
0
02,
2
0
1
5
.
[20
]
A
.
A.
A
bd
u
l
lah,
N
.
M
.
P
os
dz
i
,
a
nd
Y
.
N
i
s
h
i
o
,
“P
rel
i
min
a
ry
s
t
u
d
y
of
p
n
e
um
o
n
ia
s
ym
pto
m
s
det
ecti
on
met
h
o
d
using
Ce
ll
u
l
ar
N
eural
Network
,
”
In
tern
atio
na
l
Conf
eren
ce
on
Ele
ctri
cal
C
ontro
l
and
Com
p
u
t
er
E
n
g
i
n
eeri
n
g
2
0
1
1
InE
C
CE.
I
E
E
E,
pp
.
4
97
–50
0,
2011
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN: 2502-
4752
I
n
do
n
e
si
an
J
E
l
e
c
E
n
g
&
C
o
m
p
S
ci
, V
o
l
.
1
4
,
No. 1, April 2019 :
210 –
2
0
1
8
21
8
[
2
1
]
S
.
D
u
a
n
,
X
.
H
u
,
L
.
W
a
n
g
,
S
.
G
a
o
,
a
n
d
C
.
L
i
,
“
H
y
b
r
i
d
m
e
m
r
i
s
tor/RTD
s
t
ruct
ure-based
cell
u
l
a
r
neu
r
al
n
etw
o
rk
s
with
a
pp
li
c
a
tio
n
s
in
i
m
ag
e
pro
cess
i
ng,
”
N
e
ural
C
om
pu
t.
Ap
p
l.
,
2
01
3
.
[22]
X
.
Hu,
G.
F
eng,
S
.
Duan,
an
d
L.
L
i
u
,
“A
m
emr
i
st
i
v
e
mul
tila
y
e
r
cell
u
l
a
r
neural
n
et
wo
rk
w
it
h
ap
p
l
i
cati
o
n
s
t
o
im
ag
e
pro
c
es
si
n
g
,” IEE
E T
r
ans
.
N
eural Netwo
r
k
s
L
e
a
rn.
Sys
t
.,
v
o
l
.
2
8
,
n
o.
8
,
pp.
18
8
9
–
1
901
,
2
017.
[23
]
M
.
Di
M
a
r
co
,
M.
F
o
r
ti
,
and
L
.
P
an
cio
n
i
,
“
M
e
m
r
isto
r
stan
dar
d
cellu
lar
neural
n
etw
o
rk
s
com
puti
n
g
i
n
t
h
e
f
l
u
x
–
charg
e
d
o
m
ai
n,
”
Neural
N
e
t
wo
rk
s
,
vol.
93,
pp.
1
5
2
–1
64,
2
0
1
7
.
[24
]
Y
.
Z
hon
g,
B
.
Sh
iri
n
zadeh
,
an
d
X
.
Y
uan,
“
Optim
a
l
r
o
b
o
t
p
a
t
h
p
l
a
n
n
i
n
g
w
i
t
h
c
e
l
l
u
l
a
r
n
e
u
r
a
l
n
e
t
w
o
r
k
,
”
I
n
t
.
J
.
Int
e
ll
.
M
echat
roni
cs
R
o
bot.,
vol.
1,
n
o
.
1
,
pp.
1
8
–
3
7
,
201
1.
[2
5]
K
.
X
i
e,
Y
.
Yan
g
,
Y.
X
in,
an
d
G.
X
i
a
,
“Cel
l
u
lar
Neu
r
al
N
etw
ork-Bas
e
d
M
e
th
od
s
f
o
r
D
i
s
t
rib
u
t
e
d
Net
w
o
r
k
In
tru
s
io
n De
te
c
tio
n,”
v
o
l
.
2
01
5
,
2
01
5.
[2
6]
F
.
Al
M
a
c
ho
t
,
M
.
Ali,
A
.
Ha
j
Mosa,
C.
S
ch
wa
rzl
m
ü
l
ler,
M
.
G
ut
m
a
nn
,
an
d
K.
K
yam
a
ky
a,
“
Real
-ti
m
e
rain
dro
p
det
ectio
n
based
on
cell
u
l
a
r
neu
r
al net
wo
rk
s
f
o
r
A
D
AS
,” J. Real-
Tim
e
I
m
a
ge
P
ro
cess
.
,
p
p
.
1–1
3
,
2016
.
[27
]
K
. K
a
racs
, Á. Zarán
d
y
, P
. S
z
olg
a
y, C.
Rekeczky
, L
-Kék,
V.
Sza
b
ó
,
G
. Paz
i
e
n
za,
a
n
d
T
. Ro
s
ka,
“So
f
t
w
are
Li
brary
f
o
r
Cellul
ar W
av
e
Co
mp
ut
ing
En
g
i
nes
in
a
n
era
of
k
il
o-p
rocess
o
r ch
ip
s,” 2
0
1
0
B
I
OGRAPHIES
O
F AUTHO
RS
D
r
Azi
an A
zamim
i A
b
d
u
llah
has ob
tain
ed h
er d
e
g
ree and
m
a
st
er’s
d
e
gree
f
r
o
m
T
he
U
ni
versit
y
o
f
To
k
u
s
h
i
m
a
, Japan
in
El
ectrical
and
El
ectro
ni
c
En
gin
eerin
g
. S
he
a
ls
o
h
a
s
co
mp
leted her
d
o
cto
r
ate d
e
gree
f
rom Nara In
s
t
itu
te
o
f
S
c
ien
ce and
Tech
no
log
y
(NAIST
),
J
ap
an
i
n
201
7.
P
r
evi
o
u
s
l
y
,
she work
ed as
an
e
ngi
neer at
To
shiba El
ectronics
an
d
curren
t
l
y
s
erves
as a s
eni
o
r
l
ecturer
a
t
U
n
iv
e
r
sity
M
alay
sia
Perl
is. H
e
r researc
h
i
n
t
erest
s
are b
i
o
i
nf
orm
a
ti
cs
,
a
r
tif
i
c
i
al
i
n
t
e
lli
g
en
ce, bi
g
data
and
mach
ine learn
i
n
g
.
Mrs
Aafio
n
F
o
n
ett
a
D
i
c
kso
n
G
io
ng
h
as
o
bt
ained
her
d
e
gr
ee
f
rom
Un
iv
e
r
s
it
i
M
a
l
a
y
s
i
a
P
er
li
s
in
Bi
om
ed
i
cal E
lect
ro
ni
c En
g
i
neerin
g
prog
ram
m
e.
D
r
N
i
k
A
dilah
Hani
n
Zahri
obtained
her
d
e
gree
a
n
d
m
a
ster’s
d
egr
ee i
n
C
om
pu
ter S
c
i
e
nce an
d
M
e
di
a f
r
om
U
n
i
vers
it
y
of
Yam
anas
hi
,
J
a
pan.
S
h
e
a
l
s
o
has
com
p
let
ed h
er do
c
torate deg
ree f
r
om
t
h
e s
a
me
universit
y
i
n
M
edical
En
g
i
n
eerin
g (H
u
m
an
En
v
iro
n
m
e
n
t
M
ed
ical Eng
in
eerin
g).
Cu
rren
tl
y,
s
he
w
o
r
ks
a
s s
e
ni
or l
e
c
turer
at
S
cho
o
l
of
C
om
puter
a
n
d
Co
mmu
n
i
cati
on En
g
i
n
eerin
g
in
U
n
i
v
e
r
s
it
i
M
a
l
a
ys
ia
P
e
r
l
i
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.